Workshop: Datenvisualisierung mit ggplot2

Tag II

Author
Affiliation

Arieja Farugie & Luca Schnatz

Institut für Psychologie

Published

October 5, 2023

library(tidyverse)

exercise_data <- readRDS(url("https://github.com/jlschnatz/WoMepS_DataVizR/raw/main/clean_exercise_data.rds"))

Scales

# Erstellen eines Ridgeline-Plots
library(ggridges)
p0 <- exercise_data %>%
  drop_na(lebenszufriedenheit, subj_schichteinstufung) %>%
  ggplot(
  data = .,
  mapping = aes(x = psychological_distress,
                y = subj_schichteinstufung,
                fill = geschlecht)
  ) + 
  geom_density_ridges(scale = 0.8, 
                      alpha = 0.8,
                      rel_min_height = 0.001,
                      color = NA)

print(p0)

Achsenbeschriftungen

Möglichkeit I

Über Funktionen der Familie scale_*aesthetic*_*specification*. Was kann über diesen Familienstamm verändert werden?

  • Achsentitel
  • Achsenbreaks
  • Achsenlimits
  • Achsenbeschriftung
  • Achsenpadding
  • (und noch vieles mehr: alpha, color, fill, shape, linesize, size, etc.)


p0 +
  scale_x_continuous( # kontinuierliche x-Achse
    name = "Psychologischer Distress", # Titel der x-Achse
    limits = c(10, 60), # Achsenlimits
    breaks = seq(0, 60, 10), # Achsenabschnitte
    expand = c(0.01, 0) # Padding 
  )


p0 +
  scale_y_discrete( # diskrete y-Achse
    name = "Subjektive Schichteinstufung", # Titel der y-Achse
    labels = str_to_title, # Verändung der Labels
    expand = c(0.05, 0.05) # Padding 
  )


# Alternative:
p0 +
  scale_y_discrete( # diskrete y-Achse
    name = "Subjektive Schichteinstufung", # Titel der y-Achse
    labels = letters[1:6], # Vektorinput!
    expand = c(0.05, 0.05) # Padding 
  )

Möglichkeit II

Über die lim und lab-Familie:


p0 + 
  xlim(10, 60) + 
  xlab("Psychologischer Distress") + 
  ylab("Subjektive Schichteinstufung")


p0 + 
  labs(x = "Psychologischer Distress",
       y = "Subjektive Schichteinstufung",
       title = "Mein tolle Abbildung",
       subtitle = "Das in ein Untertitel",
       tag = "A",
       caption = "Das ist eine Beschreibung")

Titel

p0 + ggtitle(label = "Mein Titel", subtitle = "Mein Subtitel")

p0 + labs(title = "Mein Titel", subtitle = "Mein Subtitel")

Colors

Wenn wir in den Aesthetics Variablen aus dem Datensatz Parametern wie fill oder color zuordnen, können wir ebenfalls innerhalb der Familie scale_aethetic_specification die Farben ändern. Unterscheidung zwischen:

  • scale_fill_…(discrete, continuous, manual, binned, gradient, binned, brewer)
  • scale_color_…(discrete, continuous, manual, binned, gradient, binned, brewer)

Es gibt außerdem zahlreiche externe Packages, die gute Farbpaletten für sowohl qualitative und kontinuierliche Daten bereitstellen. Beispiele:

  • wesanderson
  • jcolors
  • scico
  • viridis
  • RColorBrewer
  • …

library(ggsci)
p0 +  scale_fill_manual(values = c("#3C307E", "#95CBE3"))

p0 + scale_fill_brewer(palette = "Set1")

p0 + ggsci::scale_fill_simpsons()

RColorBrewer::display.brewer.all()

Kontinuierliche Farbpaletten:


p_cor <- exercise_data %>%
  select(starts_with("ps_")) %>%
  cor(use = "complete.obs") %>%
  as.data.frame() %>%
  rownames_to_column(var = "var1") %>%
  pivot_longer(
    cols = -var1, 
    names_to = "var2",
    values_to = "pearson_cor") %>%
  mutate(across(var1:var2, ~str_extract(.x, "ps_[0-9]{2}"))) %>%
  ggplot(
    data = .,
    mapping = aes(x = var1, y = var2, fill = pearson_cor)
  ) + 
  geom_raster(alpha = 1)

print(p_cor)


p_cor + scale_fill_viridis_c(option = "C") # mit option = "A-E" andere Palette

p_cor + scico::scale_fill_scico(palette = "lajolla")

p_cor + scale_fill_gradient2(low = "red", high = "blue", mid = "white", midpoint = 0)

Themes

Vorgefertigte Themes

Es gibt verschiedene Theme-Settings, die bereits in ggplot2 als fertige Funktionen implementiert sind:


print(p0)

p0 + theme_dark()

p0 + theme_void()


library(ggthemes)

# Standard:
p0 + theme_gray()

Individuelle Anpassung

Wortstämme in den Argumenten der theme()-Funktion:

  • axis.placeholder (title, text, ticks, line)
  • legend.placeholder (title, text, background, direction, position, box, key, justification)
  • panel.placeholder (background, border, grid)
  • plot.placeholder (background, title, subtitle, caption, tag, margin)
  • strip.placeholder (background, text, clip, placement, switch)

Theme-Elements:

# Beispiel: Wie entferne ich per Hand den grauen Hintergrund und füge  Achsenlinien hinzu? 

p_axis <- p0 + theme(
  panel.background = element_rect(fill = NA),
  axis.line = element_line(color = "darkgrey", linewidth = 0.25),
  axis.ticks = element_line(color = "darkgrey", linewidth = 0.25),
)

print(p_axis)



# Beispiel: Wie passe ich die Legende an?

p_axis + theme(
  legend.position = "bottom",
  legend.title = element_text(face = "bold", size = 10),
  legend.direction = "horizontal",
  legend.key.height = unit(0.5, "cm"),
  legend.key.width = unit(0.5, "cm"),
  legend.text = element_text(size = 9)

)

Alles in Kombination:

library(sysfonts)
library(showtext)
font_add_google("Source Sans Pro", "ssp")
showtext_auto()

p0 + 
  scale_x_continuous(
    name = "Psychologischer Distress",
    limits = c(10, 60),
    breaks = seq(10, 60, 10),
    expand = c(0.04, 0.04)
  ) + 
  scale_y_discrete(
    name = "Subjektive Schichteinstufung",
    labels = str_to_title
  ) + 
  jcolors::scale_fill_jcolors(
    name = "Geschlecht",
    labels = c("Männlich", "Weiblich"),
    palette = "pal7"
    ) +
  theme(
    text = element_text(family = "ssp"),
    panel.background = element_rect(fill = NA),
    axis.line = element_line(color = "darkgrey", linewidth = 0.25),
    axis.ticks = element_line(color = "darkgrey", linewidth = 0.25),
    axis.ticks.length = unit(1, "mm"),
    axis.text.x = element_text(margin = margin(t = 5)),
    axis.text.y = element_text(margin = margin(r = 5)),
    axis.title = element_text(size = 11, face = "bold"),
    axis.title.y = element_text(margin = margin(r = 10)),
    axis.title.x = element_text(margin = margin(t = 10)),
    legend.position = c(0.92, 0.5),
    legend.title = element_text(face = "bold")
  )

Patchwork

# install.packages("patchwork")
# library(patchwork)
# library(tidyverse)
# library(here)
# 
# exercise_data <- read_rds(here("data/exercise_data/clean_exercise_data.rds"))
# glimpse(exercise_data)

Basic-Plots erstellen


p1 <- ggplot(
  data = exercise_data,
  mapping = aes(x = bmi)
  ) + 
  geom_histogram(bins = 50, fill = "lightgrey", color = "white") + 
  theme_light()

print(p1)


p2 <- ggplot(
  data = exercise_data,
  mapping = aes(x = alter, y = bmi, color = geschlecht)
) + 
  geom_point(alpha = 0.2) + 
  geom_smooth(method = "lm", fullwidth = TRUE, color = "black") + 
  facet_wrap(~geschlecht) +
  ggsci::scale_color_npg() +
  theme_light()

print(p2)


p3 <- ggplot(
  data = exercise_data,
  mapping = aes(x = geschlecht, y = bmi, color = geschlecht)
) + 
  geom_jitter(width = 0.1, alpha = 0.2) +
  stat_summary(
    fun = mean,
    geom = "pointrange",
    fun.min = function(x) mean(x) - sd(x),
    fun.max = function(x) mean(x) + sd(x),
    color = "black"
  ) + 
  ggsci::scale_color_npg() +
  theme_light()

print(p3)


p4 <- ggplot(
  data = exercise_data,
  aes(x = alter, fill = geschlecht)
  ) +
  geom_density(alpha = 0.5) +
  ggsci::scale_fill_npg() +
  theme_light()

print(p4)

Patchwork-Syntax

  • p1 + p2 um zwei Plots in horizontaler Ebene nebeneinander darzustellen (äquivalent: p1 | p2)
  • p1 / p2 um zwei Plots in der vertikalen Ebene unter/übereinander darzustellen
library(patchwork)
p1 + p2 # oder p1 | p2


p1 / p2

  • beliebige Kombination aus + bzw. | und /
  • durch Klammerschreibweise können “komplexere” Kombinationen gebildet werden

(p1 + p3) / p2

  • gemeinsame Legende:
(p1 | p3) / p2 

(p1 | p3) / p2 + plot_layout(guides = "collect")

  • Kontrolle der Breite und höhe der einzelnen Teilplots:

(p1 | p3) + plot_layout(widths  = c(2, 3))

(p3 / p4) + plot_layout(heights = c(1, 2))


(p1 + p3) / (p2 + p4) + plot_layout(widths = c(1, 2), heights = c(1, 3))


p1 + p2 + plot_annotation()